oves the phase 
image derived 
iave found to 
led and multi- 
[gorithm when 
unwrap even 
bus unwrapped 
1ality maps are 
e. 
been evaluated 
signal-to-noise 
11 unwrapping 
t higher SNR 
ective. Quality 
st performance 
quality maps 
, and yields 
to thank Dr. 
e and Water 
ntre (ISRO), 
ng, J.A., 2000. 
Aperture Radar 
83-C (12), pp. 
interferometry- 
nal Journal of 
ensional phase 
tware. Wiley- 
93p. 
Cl 1988. 
isional phase 
etry. Scientific 
1s of multigrid 
ransactions on 
-738. 
correlation in 
nsactions on 
IAPRS & SIS, Vol.34, Part 7, "Resource and Environmental Monitoring", Hyderabad, India,2002 
A SIMULATION STUDY ON RETRIEVAL OF LEAF AREA INDEX USING PRINCIPAL 
COMPONENT INVERSION TECHNIQUE 
Sasmita Chaurasia and Vinay K. Dadhwal 
Crop Inventory and Modelling Division 
Agricultural Resource Group, RESA 
Space Applications Centre, ISRO 
sasmitas(@ yahoo.com, dadhwalvk@hotmail.com 
AHMEDABAD - 380015, INDIA 
* Commission VII, WG VII/6 
KEYWORDS: Leaf Area Index (LAI), Principal Component Inversion (PCI), Canopy Reflectance Model 
ABSTRACT: 
A simulation study has been carried to investigate the use of Principal Component Inversion (PCI) technique for the retrieval of leaf 
area index (LAI). The PROSAIL model has been used for the forward analysis for the estimation of multispectral reflectance for a 
total of 2880 combinations of LAI, soil reflectance, leaf inclination angle (61), chlorophyll a+b concentration (Cab), sun and view 
angle (0s, Ov). The developed model (three separate and one combined) when tested for independent 100 samples with LAI range 
0.1-7.0 with different soil and the combination of three type of soil as background indicated that the retrieved LAI from PCI has 
higher accuracy (RMSE= 0.37, 0.26,0.24 and 0.37 for bright, medium, dark and combination of three soil, respectively) than the 
NDVI approach (RMSE=0.76, 0.93, 0.73, 0.75) 
1. INTRODUCTION 
Leaf area index (LAI) quantifies the amount of foliage area per 
unit ground surface area. It is an important parameter 
controlling . many biological-physical processes like 
evapotranspiration, photosynthesis and yield apart from its 
effect on radiation exchange with the atmosphere through its 
effect on albedo. LAI is also an input parameter for estimation 
of net primary production (NPP) (Nemry et al, 1996), in crop 
simulation models (Moulin et al, 1998) and a number of 
modeling studies related to agriculture and hydrology (Wiegand 
and Richadson, 1984, Kergoat, 1999). The accurate estimation 
of this biophysical parametr thus interests scientist world wide 
from different segment. A number of approaches ranging from 
empirical relationship of spectral indices to LAI and the CR 
based models have been used for the estimation of LAI from 
remote sensing data (Nemani et al., 1993; Cihlar et al, 2002; 
Myneni, 1997; Cihlar, 1997; Liu et al, 1999). 
The empirical models are site and sensor specific and unsuitable 
for application to large areas or in different seasons. The CR 
based approaches with the assumption of homogeneous canopy 
are more physical and rigorous and best suited for agricultural 
crops. However, the CR based techniques are not widely used 
due to the complexities involved in the inversion of the model. 
Thus there is a need for a simpler approach for the retrieval of 
LAI. . 
The simulation experiment has been set up to resemble with the 
real remote sensing data. The principal component analysis has 
been used successfully in diverse fields (Price, 1990:1992, 
Rabbett et al, 2001, Charlock et al (1990); Bess et al, 1992, 
Haskins et al, 1999). Chauhan and Nayak (1998) have reported 
the retrieval of chlorophyll from IRS-MOS data following an 
approach involving direct modeling with PCA (Krawezyk et al, 
1993), 
2. METHODS 
The study has been carried out in the following steps 
a. Use of a direct model for the simulation of canopy 
reflectance as a function of viewing and leaf 
properties. . 
b. Application of PCA and development of PCI co- 
efficients for LAI retrieval and 
c. Testing of the retrieval accuracy for an independent 
set of simulated reflectances and comparison with 
NDVI based approach. 
2.1 Direct Model 
A number of CR models have been proposed and most widely 
used for crop canopies amongst them is PROSAIL 
(PROSPECT+SAIL). The PROSAIL model includes SAIL 
(Verhoef, 1984), PROSPECT (Jacquemoud and Baret, 1990) 
and the hot spot effect. SAIL model is a turbid medium model 
with assumption of homogeneous semi-infinite medium. crop 
canopy with Lambertian reflecting leaves. The input parameters 
in this model are leaf reflectance (p;) and transmittance (ty), leaf 
area index (LAI), average leaf inclination angle (0j, soil 
reflectance (ps) and the fraction of diffused incident solar 
radiation (skyl). PROSPECT model simulates the leaf optical 
properties from visible to mid infrared as a function of only 
three variables: a parameter that counts for the leaf mesophyll 
structure (N), chlorophyll a+b concentration (C,, in ug cm?) 
and leaf water content (C, in cm). 
The PROSAIL model (version 3.01, Jacquemoud, 1993) has 
been used for simulation of the canopy spectral reflectance for 
six bands centered around 500, 595, 677.5, 800, 1707.5 and 
2187.5 nm, corresponding to LANDSAT TM for a range of 
plausible input parameters listed in table 1.